[SR-Users] private memory leak - but where?

Sebastian Damm damm at sipgate.de
Tue Sep 19 17:49:30 CEST 2017


Okay, when restarting one of those operations I did earlier obviously
made Kamailio dump all its private memory information. So I grepped
everything from PID 19134 and then let the awk line from the Wiki page
run over it. Now I'm lost interpreting the results. The top usages
were those:

    108  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from core: core/rpc_lookup.c: rpc_hash_add(100)
    166  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from core: core/pvapi.c: pv_table_add(234)
    195  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from core: core/cfg.lex: addstr(1382)
   1844  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from htable: ht_var.c: pv_parse_ht_name(141)
   1862  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from core: core/pvapi.c: pv_cache_add(272)
   1887  salm /usr/sbin/kamailio[19134]: WARNING: qm_status:
alloc'd from core: core/pvapi.c: pv_parse_format(1075)

Can anybody explain what that means? When are those memory parts allocated?

As I see, those operations seem to have something to do with htables,
right now there are absolutely no write operations, but for each
packet there's at least one read operation on a hash table to see
whether something is set for this Call-ID. All htable handling is done
from my lua script. And private pseudo variables are used pretty much
read-only as well. They might get zeroed at start of processing of the
message, but right now none of them are set again later.

Thanks in advance,
Sebastian



More information about the sr-users mailing list